PERFORMANCEGLM-5.2 → Scaleway: TTFT ↑ 20%24mPERFORMANCEDeepSeek V4 Flash → Scaleway: TTFT ↑ 35%24mPERFORMANCEMistral Medium 3.5 → Scaleway: TTFT ↓ 21%24mPERFORMANCEDeepSeek V4 Flash → Scaleway: throughput ↓ 20%24mPERFORMANCEGPT-OSS 120B → Scaleway: throughput ↑ 31%24mPERFORMANCEGPT-OSS 120B → Scaleway: TTFT ↓ 24%24mPERFORMANCEGPT-OSS 120B → Together AI: throughput ↑ 20%24mPERFORMANCEKimi K3 via OpenRouter: TTFT ↓ 49%24mPERFORMANCEGLM-5.2 via OpenRouter: reliability recovered 96.8% → 100.0%24mPERFORMANCEKimi K3 via Cortecs: throughput ↑ 184%24mPERFORMANCEKimi K3 via Cortecs: TTFT ↓ 42%24mPERFORMANCEGLM-5.2 via OpenRouter: throughput ↓ 49%24mPERFORMANCEGLM-5.2 via OpenRouter: TTFT ↑ 121%24mPERFORMANCEMiniMax M3 via OpenRouter: reliability recovered 96.8% → 100.0%24mPERFORMANCEDeepSeek V4 Pro via Cortecs: throughput ↑ 27%24mPERFORMANCEGPT-OSS 120B via OpenRouter: throughput ↓ 22%24mPERFORMANCEGLM-4.7 via OpenRouter: TTFT ↓ 78%24mPERFORMANCEGPT-OSS 20B → Groq: TTFT ↑ 24%24mPERFORMANCEKimi K3 → Together AI: TTFT ↓ 34%24mPERFORMANCEDeepSeek V4 Flash via OpenRouter: throughput ↓ 17%24mPERFORMANCELlama 3.3 70B via Cortecs: TTFT ↓ 51%24mPERFORMANCEQwen3 235B via OpenRouter: throughput ↑ 37%24mPERFORMANCEQwen3 235B via OpenRouter: TTFT ↓ 29%24mPERFORMANCEDeepSeek V4 Flash → Together AI: throughput ↑ 19%24mPERFORMANCEGLM-5.3 via OpenRouter: throughput ↓ 17%24mPERFORMANCEGPT-OSS 20B via Cortecs: throughput ↑ 28%24mPERFORMANCELlama 3.3 70B → Together AI: TTFT ↓ 33%24mPRICE~moonshotai/kimi-latest cached_input_per_mtok decreased from 0.80 to 0.2924mPRICE~moonshotai/kimi-latest output_per_mtok decreased from 13.00 to 11.3624mPRICE~moonshotai/kimi-latest input_per_mtok decreased from 0.99 to 0.7524mPRICE~deepseek/deepseek-v4-flash-latest cached_input_per_mtok decreased from 0.0077 to 0.001124mPRICE~deepseek/deepseek-v4-flash-latest output_per_mtok decreased from 1.28 to 1.0424mPRICE~deepseek/deepseek-v4-flash-latest input_per_mtok decreased from 0.0077 to 0.003824mPRICEz-ai/glm-5.1 cached_input_per_mtok increased from 0.18 to 0.2624mPRICEz-ai/glm-5.1 output_per_mtok increased from 3.03 to 4.4024mPRICEz-ai/glm-5.1 input_per_mtok increased from 0.96 to 1.4024mPRICEtencent/hy4-preview cached_input_per_mtok increased from 0.038 to 0.04224mPRICEtencent/hy4-preview input_per_mtok increased from 0.75 to 0.8324mPRICEtencent/hy3 cached_input_per_mtok increased from 0.021 to 0.03324mPRICEtencent/hy3 output_per_mtok increased from 0.33 to 0.5324m

Cerebras vs Together

3 shared models · US vs US · independent benchmarks

Registry facts

FactCerebrasTogether
Typeinference hostinference host
HeadquartersUSUS
Catalogued endpoints11
EU residencyNoNo

GPT-OSS 120B: Cerebras vs Together

● MEASURED Cerebras detail → · ● MEASURED Together detail →

MetricCerebrasTogether
TTFT p50184 ms182 ms
TTFT p95323 ms467 ms
Throughput1749 tok/s138 tok/s
Reliability100%93.5%
Tool calling100%100%
Structured output100%85.7%
$ output / M tokens$0.75$0.60

Every figure above is an independent measurement from the canonical vantage; unmeasured fields show — until measured. How we benchmark →